On Wednesday, the second day of the IMS2013 show, I had some great observations. There were lots of great questions from engineers dropping by the booth.
The general stream of traffic on the floor was much like the first day but just a bit heavier as it seemed a few more folks probably took advantage of the exhibit only badges for the day. Thus far, the IMS2013 show in Seattle had been going pretty well and the first day of the show brought a good deal of interested engineers by the booth with many great questions. Questions ranged from JESD204B to interleaved converters and back to overvoltage protection on the analog inputs of the converter.
I quite enjoy doing the show each year as it brings on good questions and great discussions. I have the time for a few days to meet face-to-face with many great engineers in the industry and talk about the problems that they are facing in their current and future designs. This is a great way to provide customers with helpful information as well as for me to get feedback and guidance of future products. It is also a great forum for feedback on the support customers receive throughout the year on the products they are using. It is great to hear that they feel very well supported.
Now, let's take a look at some of those questions. Once again, engineers were interested in some of the motivations behind the transition to JESD204B for the converter interface. Since I've touched on that already, we'll take a look at some other questions. It's coincidental that I have been blogging about interleaved converters (and have more to come on that topic by the way). That seems to be another hot topic out there among engineers using high-speed ADCs.
I had some great discussion around the advantages of interleaving as well as what the tradeoffs are. Interleaving multiple converters means that there is more than one converter used. Inevitably, there are inherent mismatches between those converters. There are mismatches in the offset, gain, timing, and bandwidth of the two converters and these have to be considered and dealt with properly. We'll not dive into that deeply here as it takes a bit more discussion to properly explain the concepts (so, another blog, another time). That said, we can move on to the questions on overvoltage protection for the analog inputs of the converter.
This is a great issue to pursue, as to my knowledge there isn't a solution on the market that can be implemented without degrading the linearity performance of the converter. The reason is that clipping devices that would act as limiters are usually diodes which are inherently nonlinear and result in harmonics that degrade the spurious free dynamic range (SFDR) performance of the converter. In many high performance systems today, there is a push to continue to improve the SFDR.
The addition of components that degrade the linearity is definitely not desired. This can be offset somewhat by placing clipping components before the anti-aliasing filter (AAF). This is much better from a linearity standpoint because the AAF can help suppress the nonlinearities generated by the clipping components. This raises a question about the response of the AAF: Is it sufficiently sharp to filter all the nonlinearities caused by the clipping components? I think the best solution is some sort of clipping component that is more linear and doesn't degrade the SFDR performance of the converter. The real question then becomes: Is such a component out there?
So once again it was a great day of questions and very good "food for thought" for both myself and the engineers who dropped by the booth. There were lots of other questions and topics discussed but these hit a lot of important areas with converters today.
These discussions help to drive new product innovations and new design concepts. I feel very rewarded to have had a good second day at the show where great ideas were discussed with fellow engineers in the industry. The first two days proved to be very good, so how was the third day? We'll look at that next time. Meanwhile, I welcome your thoughts on a method to perform more graceful clipping. Can it be done?